Less-Tap: A Fast and Easy-to-learn
Text Input Technique for Phones
Andriy Pavlovych
Wolfgang Stuerzlinger
Department of Computer Science
York University
Toronto, Ontario, Canada
{andriyp, wolfgang}@cs.yorku.ca
Abstract
A new technique to enter text using a mobile phone
keypad, Less-Tap, is described. The traditional touchtone phone keypad is ambiguous for text input because
each button encodes 3 or 4 letters. As in Multitap, our
method requires the user to press buttons repeatedly to
get a required letter. However, in Less-Tap, letters are
rearranged within each button according to their
frequency. This way, the most common letters require
only one key press.
Unlike dictionary based methods, Less-Tap facilitates
the entry of arbitrary words. Unlike LetterWise and T9®,
Less-Tap allows entering text without having to visually
verify the result, after some initial training. For English,
Less-Tap requires an average of 1.5266 keystrokes per
character (vs. 2.0342 in Multitap).
We conducted a user study to compare Less-Tap
against Multitap. Each participant had three 20-minute
sessions with each technique. The mean entry speed was
9.5% higher with the new technique.
Keywords: Text input, mobile phones, communication,
letter probabilities.
1 Introduction
Short text messaging (SMS) is very popular in the world.
According to the GSM World Association, 24 billion
messages were sent through GSM networks worldwide
during May 2002. It should be noted, however, that the
numbers are more modest for North America.
Nevertheless, it is very likely that the number of text
messaging users in North America will increase
significantly in the near future.
A major problem in using text messaging, however, is
how text can be entered on a telephone keypad.
1.1 Mobile Phone Keypad
The layout of the keypad of a telephone (Figure 1) is
standardized. The standard (ITU E.161, also known as
ANSI T1.703-1995/1999 or ISO/IEC 9995-8:1994) [8]
describes the layout of the 12 keys and the way the
letters are assigned to keys.
Mnemonic Numbers
The presence of letters on numeric keys is often used to
create a mnemonic representation of telephone numbers
(e.g. 310-BELL for 310-2355 or 1-800-5NUMBER for
1-800-568-6237). Consequently, it is a good idea to
maintain the assignment of letters to keys. However, it is
not crucial to maintain the order of letters on each key.
Figure 1: The standard 12-key keypad.
1.2 Existing Approaches
Here, we will briefly describe the existing techniques for
text entry for phone keypads.
Multitap
Most phones offer Multitap as the standard choice for
text entry. In order to enter a letter with Multitap, a user
presses a corresponding key repeatedly until the letter
appears (e.g. press ‘4’ three times to enter ‘i’, ‘7’ four
times to enter ‘s’). A notable difficulty with Multitap is
entering consecutive letters that appear on the same key
(segmentation). There are two ways to deal with this
situation. One alternative is to use a timeout after which
the system advances to the next letter. Another
alternative is to advance the cursor using a dedicated
key. The first approach requires fewer keystrokes; the
second tends to be faster for expert users, even though
the number of key presses is greater [5].
Two-key Input
As the name implies, two key presses are required for
each letter. The first press selects the group of letters
(e.g. ‘4’ selects GHI). The second press then selects the
letter from the group (e.g. ‘2’ selects ‘h’). In this
approach there is no issue of segmentation.
Another interesting two-key approach, MessagEase is
described in [7]. However, it uses a non-standard layout.
Both methods are not much faster than Multitap [9] and
are not discussed further.
Dictionary-based Disambiguation
There are several implementations that use a
disambiguation based on a dictionary: T9 from Tegic
Communications, iTap from Motorola, eZitext from Zi
Corp. All of them act very similarly – the user presses
each key only once. Once the SPACE key is pressed, the
system tries to match all possible interpretations of the
entered key sequence to the words that are contained in
the dictionary. The most probable word is the default
choice offered. If this is not the word intended by the
user, he or she has to press a special key (‘next’ key)
until the intended word appears. This means that text
entry speed goes down if a desired word is one of the
less likely choices in the dictionary.
This class of systems fails if the word is not in the
dictionary or is misspelled. In this case, the user will
have to either use a sequence of backspaces to correct
the error, or re-enter the word from the beginning using
another technique, usually Multitap.
There are additional indications that the actual process
of entering text using dictionary based methods is
somewhat distracting [2], main complaints being about
the unpredictable nature of the system’s response to the
entry of non-dictionary words. Since the user doesn’t
know if the letters that are shown to him or her on the
screen during the entry will ever be ‘magically’
converted to the intended letters, this approach has a
relatively high cognitive overhead, as the user needs to
visually verify the result.
Prefix-based Disambiguation
The only system that uses prefix-based disambiguation is
LetterWise [5]. It works by guessing the most probable
next letter based on what the user entered as the
beginning of the word – up to three consecutive letters
are analyzed in their current implementation. LetterWise
was designed to avoid the drawbacks associated with the
systems mentioned before. Since it uses a database of
prefixes (prefix is the sequence of letters preceding the
current keystroke) instead of words, the technique does
not fail when a user attempts to enter non-dictionary
words. To deal with the case when the letter that initially
appears is not the desired one the system employs a
‘next’ key, as an alternative to multiple key presses.
Like dictionary-based disambiguation methods,
LetterWise forces users to pay close attention to the
screen while entering text, to verify that the prediction of
the system matches the users intention. Therefore, eyesfree input is not possible.
Miniature QWERTY Keyboards
While the miniature keyboards seem to be a natural
choice for the text entry, their use in mobile phones is
limited, mainly due to their size and the inability to
touch type. Also, the single-handed performance for
miniature QUERTY keyboards is significantly lower
than double-handed performance.
Fastap [3]
This method uses more than 12 buttons. Each letter has a
dedicated small button, and the button for space is
double-sized. The buttons are arranged in a 4-by-7 grid
(28 ‘nodes’ and 18 ‘cells’) and the letters are assigned to
the buttons sequentially, in alphabetical order. To enter a
letter, the user simply presses the corresponding key.
We are not aware of the authors’ conducting a user
study to compare the actual performance of their method
against other alternatives.
2 Less-Tap
2.1 Motivation
The motivation behind designing Less-Tap was to design
a text entry technique for mobile phones that would be
easy to use, easy to learn, easy to implement, yet still be
compatible with standard keypad layout. Obviously, to
be of any use, the technique would have to be faster than
Multitap.
Non-dictionary Words and International Users
One interesting fact is that the language commonly used
in short text messaging is often quite different from the
language that is used to write books [5]. This poses a
major problem for dictionary-based disambiguation
methods, as they limit the user to contents of the
dictionary, more precisely to the dictionary built into the
device. Words outside a dictionary can be handled only
by switching dictionaries or by using Multitap, for
example. Another issue is the fact that ethnic minorities
in many countries often want to communicate between
themselves in their language, which may be not
Eyes-free Input
Despite its modest speed, Multitap has an advantage
over other existing systems. It allows experienced users,
even those who have not reached the expert level yet, to
enter text without looking at either the keypad or the
display. This removes the visual focus of attention from
the task and enables so-called eyes-free input [6]. Eyesfree input is very useful in certain situations such as
working in confined environments, experiencing bad
lighting conditions, exposure to excessive vibration,
wearable computing, etc. 1 The new technique presented
here, Less-Tap, is similar in that respect to Multitap, as
the result of a key press is deterministic.
2.2 Letter Frequencies in English
The arrangement of letters on the keys in our method
depends entirely on letter frequencies in English. In this
paper, that data was derived from the British National
Corpus [1].
2.3 Keystrokes per Character (KSPC) Metric
Keystrokes per character (KSPC) is a useful metric for
comparisons between two text-entry methods [4]. For
simplicity, we limit the following discussion only to
lowercase letters and ignore capitals and special
symbols. In this case, KSPC is equal to one for a
standard ten-finger keyboard since there is a dedicated
key for each letter. Given a text-entry method and a
language corpus, it is straightforward to compute a
KSPC value for that method by dividing the number of
key presses required by the number of characters in the
corpus.
The published numbers are 1.1500 for LetterWise,
1.0072 for T9, and 2.0342 for Multitap with the use of a
1
timeout kill key [5]. Less-Tap with a timeout kill key has
a KSPC value of 1.5266.
In our user tests, the timeout kill key was not used.
That would give a KSPC value of 1.9488 for Multitap
and 1.4412 for Less-Tap.
It should be noted, however, that the KSPC is not the
only parameter affecting the speed of the input. For
example, if the layout is unfamiliar and/or confusing, the
speed with which the keystrokes are performed will be
significantly lower thus affecting the final text entry
speed. We will investigate this later in the paper.
2.4 Less-Tap: a New Technique for Text Entry on
Keypads
Less-tap differs from Multitap in the order in which the
letters appear upon pressing a key. The objective was to
allow the entry of the most frequent letter on each key
with one keystroke, the second most frequent letter with
two keystrokes and so on. Figure 2 illustrates the
concept.
Figure 2: Less-Tap Layout.
By keeping the letters on the same keys, we ensure
that the keypad remains standard compliant and that the
1-800 numbers can still be entered.
The Distribution of Key Strokes
To illustrate the effect of our re-arrangement of the
letters, we had computed the distribution of the number
of key presses required to enter the complete text of our
corpus.
80%
percentage
supported by the devices that they can use. In addition,
there are some nations in the world whose population is
too small to warrant a localization of some particular
communication device. One of the examples of small
nations is Iceland, with a population of just several
hundred thousand.
In these two cases, Multitap is currently the only
option for text entry. The technique described in this
paper, Less-Tap, was designed to work best with
English. However, it should work well in the two
situations described above, possibly with a small drop in
performance. For fairness, we must acknowledge that
LetterWise also deals well with words from languages
that were not considered when building its database.
60%
Multitap
Less-Tap
40%
20%
0%
1
2
3
4
Number of keypresses per letter
This in no way implies that we, the authors, condone the use of text
messaging during driving, running, operating complex machinery
and any other activity that requires full concentration.
Figure 3: Frequency of different ‘multistrokes’.
As shown in Figure 3, Less-Tap lets the user enter
about two-thirds of all characters (65.4%) with a single
key press. The number of double presses is similar to
that of Multitap, but there are fewer triple and extremely
few (0.11%) quadruple key presses.
In the next section we will present an empirical
evaluation of Letter-Wise. As in [5], we used Multitap as
the technique to compare our technique against.
front of the slanted display beside the output window
where the text that they were entering was displayed.
Some chose to hold the handset on their laps or in the
air.
3 Test Method
3.1 Participants
There were 12 participants in the test who were recruited
through advertisements posted on the university campus.
Three participants were female, one was left-handed, and
three were frequent users of text messaging. The ages
ranged from 19 to 39 with the mean of 24.3. All but two
had extensive computer experience (five years and
more). One did not own a cell phone. Three had reported
using text messaging on the cell phone weekly or daily;
another three used it monthly. All participants were paid
$20 upon completion of the user study.
3.2 Apparatus
Hardware
Unlike similar studies that used a desktop keypad [5], we
used an actual Nokia 5190 handset. To implement our
technique we connected the keypad of the mobile phone
to the keyboard of a PC. Here we exploited the fact that
both keyboards are based on a matrix layout, i.e.
horizontal and vertical rows. We electrically connected
the 4x4 grid of the Nokia keypad to a similar part of the
PC keyboard (the region delimited by the keys ‘1’, ‘4’,
‘v’, and ‘z’) with a lightweight 8-wire cable. In effect,
each key press on the Nokia keypad entered a character
into the PC. With this, we were able to maintain the form
factor, weight, size, as well as the typical ‘feel’ of a
mobile phone. The buttons were not re-labeled to reflect
the different layouts of the letters. While re-labeling
could increase the performance for the new method, we
decided to keep the hardware modifications to the phone
to a minimum. The 5100-series models are relatively
common which should make it easy to replicate this
study. In the experiment we used only the 12 standard
keys as shown in Figure 2.
Since displaying the characters on an LCD display of
the (non-working) handset is hard to achieve for
technical reasons, we chose to display the entered text on
a Wacom LCD tablet. The advantage of using a tablet
was that it allows users to quickly find the most
convenient position for their text entry sessions. For
example, participants could hold the handset right in
Figure 4: Equipment used in the experiment.
Software
The software that we used in the experiment was written
in Java and had been used previously for text entry
experiments [5]. We modified it to interpret the
characters entered on the connected phone keypad.
Furthermore, we added the option to use either Multitap
or Less-tap (referred to as Alternative Text Entry
Method during the experiment). Figure 5 shows the user
interface.
Figure 5: Snap shot of the program window.
The first text field shows the phrase to enter and the
second field shows the text that the user has entered so
far. The software recorded all key presses and the times
when they took place. Furthermore, the software did
statistical calculations such as time to enter a phrase,
error rate, average WPM speed for the phrase, key
repeat time and so on and displayed these values at the
bottom of window for control purposes.
The buttons that don’t have letters on them were
assigned the following actions: ‘*’ – backspace, ‘1’ –
end of phrase, ‘0’ and ‘#’ – both space.
The timer started with the first typed key for each
phrase and stopped with ‘1’ so that the participants could
rest between the phrases at their discretion. They were
informed about this feature of the system.
As we were testing first time users, we did not provide
a timeout kill key in our design. This decision is based
on indications that the use of a timeout kill key has a
noticeable advantage only for expert users [9]. The
timeout was set to 1.0 s. The text cursor in the software
changes from block (‘overwrite’) to line (‘insert’)
depending on whether the timeout has expired or not. In
Multitap and Less-Tap, pressing the same button within
the timeout changes the last letter entered.
We also used a second version of the program in order
to test our technique in “eyes-free” mode. In that
version, the 2nd text field as in Figure 5 was completely
blanked out (even the cursor was not visible).
Set of Phrases
The set of phrases was the same as the one used in the
test of LetterWise [5]. It is representative of English [5].
The phrases to be entered were chosen randomly from
this file.
they pressed a correct key but a wrong number of times,
they could continue pressing it until the letter appears for
the second or third time.
There were occasional “uncorrectable” errors when
users pressed an “end of phrase” button in the middle of
the phrase. In those cases, the participants were told “not
to panic” and were simply asked to enter some
additional phrases at the end of the session.
Eyes-free sessions
In their 3rd session, the participants were asked to enter
the phrases without visual feedback (they could see the
text that they had to enter but they were not able to
observe the progress). They were also asked to try not to
look at the button assignment sheet and the buttons
themselves.
5 Results
Instructions
Before the test, the participants were given brief
instruction as to their task, how the software worked,
which keypad buttons were doing what. They were told
that they were free to hold the handset in any way they
chose during the test. They were also given the freedom
to adjust the position and the orientation of the LCD
screen.
The participants were briefly instructed on how to
enter text with Multitap. For Less-tap, it was also
explained that the sequence in which the letters appear
on the screen for that system was different from that of
Multitap. Before each session, users were encouraged to
enter a few phrases for practice and to help to transition
from the previous technique to the next.
Two small paper sheets illustrating the letter layout for
both methods were available (similar to Figure 2). These
sheets were normally placed at the bottom-left corner on
the screen.
Errors
As to errors, the participants were told to pretend that
they were typing a message to their friend so that one or
two typos per phrase were OK. In other words, we did
not force the participants to enter the phrases completely
error-free. They were told that the errors could be
corrected via using a backspace (‘*’) button. Also, if
5.1 Entry Speed
Overall, the mean entry rate was 7.15 wpm for Multitap
and 7.82 wpm for Less-Tap. In other words, Less-tap
was faster by 9.5%. The main effect for entry method
was statistically significant (F1,11 = 7.32, p < 0.05).
Figure 6 demonstrates the entry speed in words per
minute for different sessions.
10
Words per Minute
4 Procedure
Each participant was assigned three time slots, 30–40
minutes each, with short breaks between them. Each slot
contained one session with each method, 20 phrases
each. We used a within-subjects design and
counterbalanced the order of entry methods to
compensate for learning effects.
8
6
Multitap
4
Less-Tap
2
0
Session 1
Session 2
Eyes-free
Figure 6: Entry speed (wpm) by entry method and
session.
5.2 Error Rate
Over all sessions, the difference in error rate was not
statistically significant between Multitap and Less-Tap
(F1,11 = 4.66, p > 0.05). However, it was significant for
the eyes-free session (F1,11 = 4.91, p < 0.05). Figure 7
shows the graphs for error rate for each session.
5.3 Errors: Unnecessary Key Presses
To analyze the errors further, we computed the number
of unnecessary key presses. This occurs, when the user
presses a key too often. Then they have to press them a
few more times to get the intended character. While it
7
6
5
4
3
2
1
0
Multitap
Less-Tap
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
Multitap
Less-Tap
Session 1
Session 1
Session 2
Eyes-free
Figure 7: Error rate vs. entry method and session.
The data for following figure was obtained by
analyzing the key press log data. The absolute number of
errors was about the same for both methods. What
differed was the distribution of the errors among
different types of key presses (single, double etc.). To
illustrate the results, we computed the percentage of
extra key presses in the total number of key presses of
each type. The resulting graph is shown in Figure 8. For
example, if 2700 key presses were required to enter all
letters that need two key presses each, and 3000 were
actually performed, than the percentage of extra presses
would be 10%.
Portion of extra
keystrokes in the total
statistically significant (F1,11 = 53.97, p < 0.05). Figure 9
shows the graphs by session.
Keystrokes per second
Error Rate, %
was possible to correct those errors with a backspace
button, almost all users preferred to continue pressing
the same button thus going for another round (e.g. the
letter ‘u’ can be entered by pressing the button ‘8’ two,
five, eight or more times).
3.5%
3.0%
2.5%
2.0%
1.5%
1.0%
0.5%
0.0%
Less-Tap
Multitap
1
2
3
Keystrokes per character
4
Figure 8: Total number of errors vs. total number of key
presses of that type.
The graph shows that the number of errors is
significantly greater for single key presses in Less-Tap.
We have no real good explanation for this, but it is likely
to be related to the distribution of the types of key
presses in both systems (Figure 3).
5.4 Key Repeat Rate
The key repeat rate, i.e. how fast keys were hit, was
lower for Less-Tap (1.04 vs. 1.27 keystrokes per
second). The main effect for entry method was
Session 2
Eyes-free
Figure 9: Key repeat rate in different sessions.
As far as we can determine, the fact that Less-Tap has
always fewer keystrokes per second (KSPS) than
Multitap can be attributed to cognitive overhead for
unusual letter sequences (i.e. ACB instead of ABC).
5.5 Keystrokes per Character
The number of keystrokes per character was highly
significant between methods (F1,11 = 332.74, p < 0.001).
Session
Multitap Less-Tap
1
2.1904
1.6254
2
2.1927
1.6808
3 (eyes-free) 2.0728
1.5660
All
2.1505
1.6215
Predicted
1.9488
1.4412
Table 1. KSPC data by session.
For both methods, the values were 10–15% greater
than those predicted. Overall, Multitap required about
33% more keystrokes, which (almost) matches the
prediction (35%).
5.6 Subjective Responses of Participants
After the test, the users were asked to fill out a short
questionnaire that contained several questions that were
rated on a seven-point Likert scale. For example, the
choices for perceived speed were: ’much faster’, ‘faster’,
‘slightly faster’, ‘no difference’, ‘slightly slower’,
‘slower’, ‘much slower’.
Summarizing the responses, in comparison to Multitap
users felt that the new system was:
• faster
• slightly easier to use
• slightly less tiring
• slightly easier to learn
• somewhat preferred, in that they wanted Less-Tap as
a permanent replacement of Multitap in their phone.
Most commented that they would prefer Less-Tap
more if the buttons were re-labeled to reflect the
changed ordering.
6 Discussion
6.1 Entry Speed vs. Sessions
It can be seen that the speed improves as the time goes
on. That can be seen from both the improvement of key
repeat time (Figure 9) and the improvement of text entry
speed in WPM (Figure 6).
Different people get tired at different times. While
some can perform a repetitive task for hours, others get
bored after 20 minutes. In the experiment, we asked
participants to work in one-hour slots. The analysis of
the log files indicated that by the end of their respective
time slot, the speed dropped for some of the users.
6.2 Why is the Improvement Lower than Predicted?
Since the predicted difference in KSPC values between
Multitap and Less-Tap is 35.2%, the question arises as to
why the improvement the text entry speed is only 9.5%?
The speed of the text entry in words per minute is
composed of the speed with which the key presses are
made, known as a key repeat rate (in keystrokes per
second), and the average number of keystrokes required
to enter character. In this paper, ‘word’ is defined as five
characters, a convention commonly used in the analysis
of text entry methods.
The results above show that the discrepancy between
the actual and predicted values for KSPC is similar for
both techniques. Now, let us go back to the Figure 9. As
we can observe, the key repeat rate is significantly lower
for Multitap than it is for Less-Tap. It took about 20%
longer to perform a button press in Less-Tap. We
attribute some of this difference to cognitive overhead.
Cognitive Overhead
Cognitive overhead is usually described as the additional
effort or concentration in keeping track of several things
(such as positions of letters on keys) at a time.
While we expected Less-Tap to be about as easy to
learn as Multitap, obviously there were some differences
which caused the users to be slower in learning Less-Tap
or using it (even though they thought the opposite and
actually showed higher entry rates). We attribute this to
additional mental processing required by Less-Tap.
Considering that, overall, participants found Less-Tap
“easier to use” and to be “slightly less tiring”, we believe
that the increased cognitive overhead caused less fatigue
than more button presses.
Key Repeat Time for Less-Tap and that of
LetterWise
It would have been interesting to compare the findings
mentioned above to those of alternative systems,
notably, LetterWise [5]. While there is no data that states
a key repeat rate for LetterWise, there is a very detailed
graph [5, Figure 6] showing the entry speed in WPM by
session for both Multitap and LetterWise. The authors of
that paper did a longitudinal study (10 hours with each
system) and they used a desktop size keypad. We
estimated from their graph the values for the 3rd session
(approx. 1.5 hrs into the test vs. 1 hr in our case). The
values obtained were 9.1 and 10.5 WPM for Multitap
and LetterWise respectively. That is, LetterWise is about
15% faster than Multitap after 1.5 hours of use.
However, the difference is much larger in terms of
KSPC: 2.03 vs. 1.15 respectively (about 77%). This
means that, with LetterWise, it takes about 50% longer
to perform key presses than with Multitap. Since LessTap required only about 20% more time for each key
press than Multitap (see Figure 9), we may conclude that
the cognitive overhead of our system is smaller that that
of LetterWise. We could find no similar data on T9.
6.3 Presence of Two Distinct Groups of Users
When analyzing the results we discovered that there
seem to be two distinct groups of participants, each
accounting for exactly half the participants. In one
group, Less-Tap is much faster than Multitap, almost
20% (8.54 wpm vs. 7.12 wpm), with high statistical
significance (F1,5 = 83.87, p < 0.001). In the other group,
the two methods were essentially the same in terms of
entry speed and error rate (for entry speed, F1,5 = 0.13, p
>> 0.05). In the second group, three users performed
slower with Less-Tap but only for one of them was the
difference greater than 2% and in that case it was caused
by a poor performance of Less-Tap during the “eyesfree” session.
We speculate that some people can easily adapt to reordered letter sequences, while others find them
significantly harder to memorize and take much longer
to learn them. However, we believe that with more use
eventually all people will adapt to the new system and
benefit from it. A long-term study would be needed to
analyze this further.
6.4 Performance in the Eyes-free Mode
It can be seen from the figures above that, in the eyesfree mode, Less-Tap performs no worse than Multitap.
And, in fact, users make fewer errors with Less-Tap in
eyes-free mode (Figure 7). Of course, some more
practice is required in order for the error rate to drop to a
tolerable level.
6.5 Comparison with Other Techniques
There are several methods to enter text on the mobile
phone keypad. None of them is perfect, in our opinion
(including our own). In the following table, we will try
to briefly summarize the advantages and disadvantages
of different techniques.
Method
Multitap
Less-Tap
LetterWise
T9
Property
Available
on mobile
phones
Enforces
spelling
Nondictionary
words
Eyes-free
input
Implement
ation*
KSPC
Almost all
None
None
known
Many newer
models
No
No
No
Yes
Yes
Yes
Yes
No
Yes
Yes
No
No
Very easy
Very easy
Moderate
Hard
more than 9.5% faster. In several cases we were able to
observe an over 30% speedup.
Less-Tap is very easy to implement, like Multitap. It
also allows for eyes-free input and does not depend on a
dictionary.
Acknowledgements
The authors would like to thank Scott MacKenzie for
providing us with software for the experiments and
William Soukoreff for his valuable insights on how to
conduct user studies, as well as NSERC for funding.
References
[1] British National Corpus, ftp://ftp.itri.bton.ac.uk/bnc.
1.4412 or
1.1500
1.0072*2
1.5266 (see
text)
Nothing or Nothing or Must change
change
change
dictionary
letter
database
ordering
* – includes memory and processing requirements
*2 – may be much greater depending on the corpus
1.9488 or
2.0342 (see
text)
Adaptation Do nothing
to other
languages
Table 2. Comparison between different text entry
methods.
6.6 Future Directions
In this paper, we did not test our technique against
another popular technique – T9. Since the two methods
work very differently, it’s hard to speculate how LessTap would compare to T9 in the real life.
Also, it should be noted that the speed at which key
presses were made was significantly lower than
theoretically possible (see [9], for example) – in both
systems. Thus, it is likely that, with further training, the
advantage of Less-Tap over Multitap will increase.
7 Conclusion
We presented a new technique to enter text on mobile
phone keypads, which requires 25% fewer keystrokes
compared to Multitap. The results of the user study show
that for first-time users the new technique is on average
[2] Gutowitz, H. Barriers to Adoption of DictionaryBased Text-Entry Methods: A Field Study.
Available at http://www.eatoni.com/research.
[3] Levy, David. The Fastap Keypad and Pervasive
Computing. In Proc. First International Conference,
Pervasive 2002, LNCS 2414, pp. 58-68, Zürich,
Switzerland, 2002.
[4] MacKenzie, I. S. KSPC (keystrokes per character)
as a characteristic of text entry techniques. To
appear in Proc. Fourth International Symp. on
Human-Computer Interaction with Mobile Devices.
Heidelberg, Germany: Springer-Verlag.
[5] MacKenzie, I. S., Kober, H., Smith, D., Jones, T., &
Skepner, E. (2001). LetterWise: Prefix-based
disambiguation for mobile text input. Proc. ACM
Symp. on User Interface Software and Technology –
UIST 2001, pp. 111-120. New York: ACM.
[6] MacKenzie, I. S., & Soukoreff, R. W. Text entry for
mobile computing: Models and methods, theory and
practice. To appear in Human-Computer
Interaction.
[7] Nesbat, S. B. Fast, Full Text Entry Using a Physical
or virtual 12-Button Keypad. Available at
http://www.exideas.com/ME/whitepaper.pdf.
[8] Phone keypads.
http://www.dialabc.com/motion/keypads.html
[9] Silfverberg, M., MacKenzie, I. S., & Korhonen, P.
(2000). Predicting text entry speeds on mobile
phones. Proc. ACM Conference on Human Factors
in Computing Systems – CHI 2000, pp. 9-16. New
York: ACM.